INDEX

Explanations

terms related to forceful actions or coercive measures

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 39-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_39/width_16k/average_l0_131

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.39.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Datuak

-0.50

exitRule

-0.43

 plomo

-0.39

apunov

-0.39

UserScript

-0.39

 Lard

-0.38

üche

-0.36

 magnétique

-0.36

 McDowell

-0.35

Scorecard

-0.35

POSITIVE LOGITS

 spray

2.42

 spraying

2.20

 Spray

2.16

spray

2.09

 sprayed

2.06

Spray

2.06

 sprays

2.06

 sprayer

1.80

喷

1.53

SPR

1.52

Activations Density 0.647%

terms related to forceful actions or coercive measures

No Comments

No Known Activations